There is a version of the Instacart success story that gets told as a lucky timing story. Right place, right time, pandemic hit, grocery delivery exploded, easy money.
That version skips the part where the founder spent two years failing at twenty different companies before landing on the idea. It skips the part where he missed the Y Combinator deadline and had to deliver a six-pack of beer to a partner’s office just to get a meeting. It skips the part where Amazon bought his biggest retail partner and every investor in the Valley sent him a text asking if the company was done.
None of those things sound like luck. They sound like a person who kept going long after a reasonable person would have stopped.
The Engineer With a Empty Fridge and Twenty Failures
Apoorva Mehta was born in Jodhpur, India, moved to Libya as a kid, then to Hamilton, Ontario at 14. He studied electrical engineering at Waterloo, went through stints at BlackBerry, Qualcomm, and eventually Amazon, where he worked as a supply chain engineer on fulfillment systems. That last job turned out to matter a lot.
In 2010, he quit Amazon, moved to San Francisco, and decided he was going to build a startup. He had no clear idea what it would be. Just the conviction that he was going to build something.
Over the next two years he tried around twenty different ideas. A social network for lawyers. A platform to book office hours with famous tech people. An ad platform for gaming. Nothing worked. Each one failed fast enough that he could move to the next without running out of money, but slowly enough that he learned something from each one. The twenty failures are not incidental to the story. They are probably a significant part of why the twenty-first worked.
The Instacart idea came from a personal situation that was almost too simple. He was sitting in his San Francisco apartment in 2012 with nothing in the fridge except hot sauce. He couldn’t cook a meal from hot sauce. He wanted groceries. And he noticed, in the way that someone who has been hunting for a startup idea for two years notices things, that in 2012 you could buy almost anything online except groceries.
He started coding. Three weeks later the first version of the app existed. He placed an order, walked to the grocery store, bought the items himself, and delivered them to himself. First customer and first shopper, same person. He gave himself a five-star review.
The Beer That Got Him Into YC
When Mehta was ready to apply to Y Combinator he missed the application deadline. Standard procedure at this point would be to wait for the next batch.
Instead, he used the Instacart app to order a six-pack of beer from the 21st Amendment Brewery and have it delivered to a YC partner’s office. The product demo was the pitch. Look, here is the app, here is a beer, here is how it works.
It worked. He got the meeting. He got into the batch.
Paul Graham’s version of startup advice has always been that the best founders show rather than tell. Mehta showed. YC helped him raise $2.3M and introduced him to his two co-founders, Max Mullen and Brandon Leonardo, who he met during the program.
The early days were genuinely scrappy in ways that are easy to romanticize but were probably extremely stressful to live. No grocery retailer would partner with them at first. They couldn’t build a real product catalog without the retailers’ data. So they went to stores, picked up one of every single item on the shelves, transported everything to a photo studio, photographed it all, and manually uploaded the catalog. A $50,000 investment in photography before a single partnership existed. When they finally cataloged their first store completely, demand at that store doubled overnight.
To solve the supply side before they had shoppers, they invented what they called “ninja shopping.” A team member would go to a store, buy the items off the shelf at retail price, deliver them to the customer, and eat the margin. Technically they were losing money on every order. But they were proving the demand was real, learning how the operations actually worked at ground level, and generating the data they needed to pitch retailers properly.
Mehta himself would take Uber to stores when there weren’t enough shoppers available. The CEO, on his way to becoming a billionaire, riding Ubers to buy other people’s groceries in 2012.
The Asset-Light Bet That Everyone Hated
Instacart launched in a shadow that was impossible to ignore. A decade before they started, a company called Webvan had raised over a billion dollars to solve the exact same problem. Webvan built warehouses. It bought fleets of trucks. It hired thousands of employees. It was the most expensive way possible to deliver groceries and it collapsed spectacularly in 2001, becoming one of the canonical cautionary tales of the dot-com era.
When Mehta walked into investor meetings in 2012, the Webvan ghost was in the room every time. At one early pitch, he titled his deck “Webvan done right” to confront it directly. An investor got up and left on the third slide.
The key insight that made Instacart different from Webvan was about asset ownership. Webvan built infrastructure. Instacart built a layer on top of infrastructure that already existed. Grocery stores were already there. Their inventory was already stocked. Their shelves were already organized. Instacart’s job was to put a digital interface on top of that existing physical network and use independent contractors to do the picking and delivery.
No warehouses. No trucks. No owned inventory. The capital efficiency of the model was the entire point. Webvan had to build everything before they could make a single delivery. Instacart could launch in a new city in days by signing up a handful of shoppers and partnering with local stores.
The gig economy framing was not accidental. By structuring shoppers as independent contractors rather than employees, Instacart kept its labor costs variable rather than fixed. When demand went up, more shoppers signed up for shifts. When demand went down, the cost structure contracted automatically. This is not a neutral observation about worker classification, there are legitimate debates about what this model means for the people doing the work, but as a financial architecture for scaling a marketplace it was dramatically more efficient than alternatives.
When Amazon Tried to Kill Them
In June 2017, Amazon announced it was acquiring Whole Foods for $13.7 billion.
Whole Foods was Instacart’s single largest retail partner at the time. Amazon was already building competing delivery infrastructure. This was not a minor competitive development. This was the biggest player in e-commerce acquiring the asset that had been most central to Instacart’s early growth.
Mehta described his reaction in an interview later. He said it was not in his risk bingo card. He got a short call, didn’t know what to say, started refreshing his feed. Investors and people’s parents were texting to ask if the company was okay. He said it felt like this was going to be his twenty-first failure.
What he did next is the most interesting strategic moment in the company’s history. Rather than finding a way to coexist with Amazon or concede the Whole Foods business gracefully, he went to war.
He put the company in what he called “war mode” and executed a high-risk plan. Instacart would sign every major grocery retailer in North America onto the platform before Amazon could consolidate. The fear that Amazon was coming for grocery was real across every major chain. Kroger, Costco, Albertsons, Walmart, Wegmans. All of them were watching what had just happened to Whole Foods and thinking about their own vulnerability.
Instacart went to them one by one with a proposition: we are not your competitor. We are your technology partner. We help you reach customers you would otherwise lose to Amazon. We drive incremental demand, not substitutional. And if Amazon is the threat, you need us more now than you did before.
It worked. In the months after the Whole Foods acquisition, Instacart signed more major retailers than it had in the previous three years combined. The Amazon threat that was supposed to end the company became the forcing function that made the partnership business dramatically larger.
When Whole Foods eventually ended the Instacart partnership in 2019 to consolidate fully with Amazon’s delivery infrastructure, Instacart barely noticed. They had 200 other retail partners by then.
The Pandemic and the $39 Billion Peak
In March 2020, grocery delivery went from a convenience to an essential service overnight.
Instacart’s sales grew 590% in 2020. The company onboarded 300,000 new shoppers between mid-March and mid-April alone to handle the surge in demand. At the peak of the pandemic, Instacart was signing partnerships, adding shoppers, and shipping orders at a pace that would have seemed impossible six months earlier.
The valuation went to $39 billion. At that number, Instacart was being valued higher than most traditional grocery chains that had been running physical stores for decades.
Two things were happening simultaneously. The genuine structural shift where millions of households tried grocery delivery for the first time and discovered they liked it, and the pandemic premium that inflated the valuations of anything related to staying home. When the pandemic premium started unwinding in 2021 and 2022, Instacart had to figure out which of those two things it actually was.
The answer, it turned out, was mostly the first one. Grocery delivery adoption did not revert to 2019 levels when people were allowed back in stores. The convenience had been established. The behavior had been learned. The numbers came down from peak but settled at a permanently higher baseline than before.
The IPO and the Advertising Pivot
Instacart went public in September 2023 at a $10 billion valuation. The peak had been $39 billion in 2021. The 75% drawdown in private valuation before going public was painful to acknowledge but honest. The company priced at $30 per share, opened above $40 on the first day, and has traded in a range since.
The more interesting story than the valuation is what the company actually was by the time it went public. Because Instacart in 2023 was not primarily a grocery delivery company. It was building toward being a retail media and commerce technology platform.
Advertising had become the highest-margin part of the business. Consumer packaged goods brands, the Coca-Colas and Unilevers and Procter and Gambles of the world, were spending money to appear at the top of Instacart search results the same way they had historically paid for prominent shelf placement in physical stores. Retail media, the term for digital advertising sold by retailers or retail platforms, was growing fast across the whole industry. Instacart had the transaction data, the audience, and the grocery context that made their advertising inventory valuable. Advertising and other revenue totaled $871 million in fiscal 2023, growing 18% year over year, and was expanding faster than the delivery business.
The Caper Cart, a smart shopping cart with a built-in screen that can scan items as customers put them in the cart, let them skip the checkout line, and show them digital promotions, was the most visible piece of the in-store technology push. Instacart was selling the technology to grocery retailers as part of a broader suite of enterprise software. If they could become the operating system that powers grocery retail both online and inside stores, the addressable market got much bigger and the dependency on delivery margins went down.
The thesis was: we have the data, the retailer relationships, and the consumer reach. Every dollar a CPG brand spends trying to influence grocery purchasing behavior is a potential Instacart revenue stream.
What Actually Made It Work
The Instacart success story is actually three separate bets that all had to go right.
The first bet was the asset-light model. Doing what Webvan tried but without owning anything. Using existing stores, existing inventory, and a gig workforce. This worked because it kept the capital requirements low enough that the company could survive its early years without perfect unit economics, and because it made geographic expansion fast and cheap.
The second bet was the retailer partnership model. Rather than competing with grocery chains by building their own supply chain, Instacart made the retailers themselves the product. The stores were the network. Every major retailer that signed with Instacart brought its customer base into the platform. The Amazon acquisition of Whole Foods, which should have been a catastrophe, became the catalyst that signed the rest of the industry.
The third bet was that delivery was the top of the funnel, not the whole business. The transaction data from 900 million orders is worth something beyond the margin on each delivery. The relationship with 9,000 CPG brands that want to reach grocery customers is worth something. The technology that powers a retailer’s online ordering and in-store operations is worth something. Delivery was how Instacart built the relationships. The data and advertising business is how it monetized them.
Mehta started with a hot sauce problem in his apartment in 2012. The company he built processes over $29 billion in annual grocery transactions and reaches 95% of US households.
The twenty failed startups before it probably mattered more than anyone gives them credit for.
Leave a Reply